1 /*=========================================================================
4 Module: $RCSfile: gdcmDocument.cxx,v $
6 Date: $Date: 2004/07/19 05:57:20 $
7 Version: $Revision: 1.53 $
9 Copyright (c) CREATIS (Centre de Recherche et d'Applications en Traitement de
10 l'Image). All rights reserved. See Doc/License.txt or
11 http://www.creatis.insa-lyon.fr/Public/Gdcm/License.htm for details.
13 This software is distributed WITHOUT ANY WARRANTY; without even
14 the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15 PURPOSE. See the above copyright notices for more information.
17 =========================================================================*/
19 #include "gdcmDocument.h"
20 #include "gdcmValEntry.h"
21 #include "gdcmBinEntry.h"
22 #include "gdcmSeqEntry.h"
24 #include "gdcmGlobal.h"
26 #include "gdcmDebug.h"
35 #include <netinet/in.h>
40 // Implicit VR Little Endian
41 #define UI1_2_840_10008_1_2 "1.2.840.10008.1.2"
42 // Explicit VR Little Endian
43 #define UI1_2_840_10008_1_2_1 "1.2.840.10008.1.2.1"
44 // Deflated Explicit VR Little Endian
45 #define UI1_2_840_10008_1_2_1_99 "1.2.840.10008.1.2.1.99"
46 // Explicit VR Big Endian
47 #define UI1_2_840_10008_1_2_2 "1.2.840.10008.1.2.2"
48 // JPEG Baseline (Process 1)
49 #define UI1_2_840_10008_1_2_4_50 "1.2.840.10008.1.2.4.50"
50 // JPEG Extended (Process 2 & 4)
51 #define UI1_2_840_10008_1_2_4_51 "1.2.840.10008.1.2.4.51"
52 // JPEG Extended (Process 3 & 5)
53 #define UI1_2_840_10008_1_2_4_52 "1.2.840.10008.1.2.4.52"
54 // JPEG Spectral Selection, Non-Hierarchical (Process 6 & 8)
55 #define UI1_2_840_10008_1_2_4_53 "1.2.840.10008.1.2.4.53"
56 // JPEG Full Progression, Non-Hierarchical (Process 10 & 12)
57 #define UI1_2_840_10008_1_2_4_55 "1.2.840.10008.1.2.4.55"
58 // JPEG Lossless, Non-Hierarchical (Process 14)
59 #define UI1_2_840_10008_1_2_4_57 "1.2.840.10008.1.2.4.57"
60 // JPEG Lossless, Hierarchical, First-Order Prediction (Process 14,
61 // [Selection Value 1])
62 #define UI1_2_840_10008_1_2_4_70 "1.2.840.10008.1.2.4.70"
64 #define UI1_2_840_10008_1_2_4_90 "1.2.840.10008.1.2.4.90"
66 #define UI1_2_840_10008_1_2_4_91 "1.2.840.10008.1.2.4.91"
68 #define UI1_2_840_10008_1_2_5 "1.2.840.10008.1.2.5"
69 // UI1_1_2_840_10008_1_2_5
70 #define str2num(str, typeNum) *((typeNum *)(str))
72 //-----------------------------------------------------------------------------
73 // Refer to gdcmDocument::CheckSwap()
74 const unsigned int gdcmDocument::HEADER_LENGTH_TO_READ = 256;
76 // Refer to gdcmDocument::SetMaxSizeLoadEntry()
77 const unsigned int gdcmDocument::MAX_SIZE_LOAD_ELEMENT_VALUE = 0xfff; // 4096
78 const unsigned int gdcmDocument::MAX_SIZE_PRINT_ELEMENT_VALUE = 0x7fffffff;
80 //-----------------------------------------------------------------------------
81 // Constructor / Destructor
85 * @param inFilename file to be opened for parsing
86 * @param exception_on_error whether we throw an exception or not
87 * @param enable_sequences = true to allow the header
88 * to be parsed *inside* the SeQuences,
89 * when they have an actual length
90 * \warning enable_sequences *has to be* true for reading PAPYRUS 3.0 files
91 * @param ignore_shadow to allow skipping the shadow elements,
92 * to save memory space.
93 * \warning The TRUE value for this param has to be used
94 * with a FALSE value for the 'enable_sequence' param.
95 * ('public elements' may be embedded in 'shadow Sequences')
97 gdcmDocument::gdcmDocument( std::string const & filename,
98 bool exception_on_error,
99 bool enable_sequences,
103 IgnoreShadow = ignore_shadow;
104 //EnableSequences=enable_sequences;
105 (void)enable_sequences;
106 EnableSequences = true; // JPR // TODO : remove params out of the constructor
107 SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
111 if ( !OpenFile(exception_on_error))
116 dbg.Verbose(0, "gdcmDocument::gdcmDocument: starting parsing of file: ",
120 fseek(Fp,0L,SEEK_END);
121 long lgt = ftell(Fp);
125 long beg = ftell(Fp);
130 long l = ParseDES( this, beg, lgt, false); // le Load sera fait a la volee
131 (void)l; //is l used anywhere ?
135 // Load 'non string' values
137 std::string PhotometricInterpretation = GetEntryByNumber(0x0028,0x0004);
138 if( PhotometricInterpretation == "PALETTE COLOR " )
140 LoadEntryVoidArea(0x0028,0x1200); // gray LUT
141 LoadEntryVoidArea(0x0028,0x1201); // R LUT
142 LoadEntryVoidArea(0x0028,0x1202); // G LUT
143 LoadEntryVoidArea(0x0028,0x1203); // B LUT
145 LoadEntryVoidArea(0x0028,0x1221); // Segmented Red Palette Color LUT Data
146 LoadEntryVoidArea(0x0028,0x1222); // Segmented Green Palette Color LUT Data
147 LoadEntryVoidArea(0x0028,0x1223); // Segmented Blue Palette Color LUT Data
149 //FIXME later : how to use it?
150 LoadEntryVoidArea(0x0028,0x3006); //LUT Data (CTX dependent)
154 // --------------------------------------------------------------
155 // Special Patch to allow gdcm to read ACR-LibIDO formated images
157 // if recognition code tells us we deal with a LibIDO image
158 // we switch lineNumber and columnNumber
161 RecCode = GetEntryByNumber(0x0008, 0x0010); // recognition code
162 if (RecCode == "ACRNEMA_LIBIDO_1.1" ||
163 RecCode == "CANRME_AILIBOD1_1." ) // for brain-damaged softwares
164 // with "little-endian strings"
166 Filetype = gdcmACR_LIBIDO;
167 std::string rows = GetEntryByNumber(0x0028, 0x0010);
168 std::string columns = GetEntryByNumber(0x0028, 0x0011);
169 SetEntryByNumber(columns, 0x0028, 0x0010);
170 SetEntryByNumber(rows , 0x0028, 0x0011);
172 // ----------------- End of Special Patch ----------------
174 PrintLevel = 1; // 'Medium' print level by default
179 * @param exception_on_error
181 gdcmDocument::gdcmDocument(bool exception_on_error)
184 (void)exception_on_error;
185 //EnableSequences=0; // ?!? JPR
187 SetMaxSizeLoadEntry(MAX_SIZE_LOAD_ELEMENT_VALUE);
190 PrintLevel = 1; // 'Medium' print level by default
194 * \brief Canonical destructor.
196 gdcmDocument::~gdcmDocument ()
201 // Recursive clean up of sequences
202 for (TagDocEntryHT::const_iterator it = TagHT.begin();
203 it != TagHT.end(); ++it )
210 //-----------------------------------------------------------------------------
214 * \brief Prints The Dict Entries of THE public Dicom Dictionary
217 void gdcmDocument::PrintPubDict(std::ostream & os)
219 RefPubDict->Print(os);
223 * \brief Prints The Dict Entries of THE shadow Dicom Dictionary
226 void gdcmDocument::PrintShaDict(std::ostream & os)
228 RefShaDict->Print(os);
231 //-----------------------------------------------------------------------------
234 * \brief Get the public dictionary used
236 gdcmDict *gdcmDocument::GetPubDict()
242 * \brief Get the shadow dictionary used
244 gdcmDict *gdcmDocument::GetShaDict()
250 * \brief Set the shadow dictionary used
251 * \param dict dictionary to use in shadow
253 bool gdcmDocument::SetShaDict(gdcmDict *dict)
260 * \brief Set the shadow dictionary used
261 * \param dictName name of the dictionary to use in shadow
263 bool gdcmDocument::SetShaDict(DictKey dictName)
265 RefShaDict = gdcmGlobal::GetDicts()->GetDict(dictName);
270 * \brief This predicate, based on hopefully reasonable heuristics,
271 * decides whether or not the current gdcmDocument was properly parsed
272 * and contains the mandatory information for being considered as
273 * a well formed and usable Dicom/Acr File.
274 * @return true when gdcmDocument is the one of a reasonable Dicom/Acr file,
277 bool gdcmDocument::IsReadable()
279 if( Filetype == gdcmUnknown)
281 std::cout << " gdcmDocument::IsReadable: Filetype " << Filetype
282 << " " << "gdcmUnknown " << gdcmUnknown << std::endl; //JPR
283 dbg.Verbose(0, "gdcmDocument::IsReadable: wrong filetype");
289 dbg.Verbose(0, "gdcmDocument::IsReadable: no tags in internal"
299 * \brief Internal function that checks whether the Transfer Syntax given
300 * as argument is the one present in the current document.
301 * @param SyntaxToCheck The transfert syntax we need to check against.
302 * @return True when SyntaxToCheck corresponds to the Transfer Syntax of
303 * the current document. False either when the document contains
304 * no Transfer Syntax, or when the Tranfer Syntaxes don't match.
306 bool gdcmDocument::IsGivenTransferSyntax(std::string const & syntaxToCheck)
308 gdcmDocEntry *entry = GetDocEntryByNumber(0x0002, 0x0010);
314 // The entry might be present but not loaded (parsing and loading
315 // happen at differente stages): try loading and proceed with check...
316 LoadDocEntrySafe(entry);
317 if (gdcmValEntry* valEntry = dynamic_cast< gdcmValEntry* >(entry) )
319 std::string transfer = valEntry->GetValue();
321 // The actual transfer (as read from disk) might be padded. We
322 // first need to remove the potential padding. We can make the
323 // weak assumption that padding was not executed with digits...
324 while ( ! isdigit(transfer[transfer.length()-1]) )
326 transfer.erase(transfer.length()-1, 1);
328 if ( transfer == syntaxToCheck )
329 //if( transfer.find( syntaxToCheck ) ) //should be faster
338 * \brief Determines if the Transfer Syntax of the present document
339 * corresponds to a Implicit Value Representation of
341 * \sa \ref gdcmDocument::IsGivenTransferSyntax.
342 * @return True when ImplicitVRLittleEndian found. False in all other cases.
344 bool gdcmDocument::IsImplicitVRLittleEndianTransferSyntax()
346 return IsGivenTransferSyntax(UI1_2_840_10008_1_2);
350 * \brief Determines if the Transfer Syntax was already encountered
351 * and if it corresponds to a ExplicitVRLittleEndian one.
352 * @return True when ExplicitVRLittleEndian found. False in all other cases.
354 bool gdcmDocument::IsExplicitVRLittleEndianTransferSyntax()
356 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_1);
360 * \brief Determines if the Transfer Syntax was already encountered
361 * and if it corresponds to a DeflatedExplicitVRLittleEndian one.
362 * @return True when DeflatedExplicitVRLittleEndian found. False in all other cases.
364 bool gdcmDocument::IsDeflatedExplicitVRLittleEndianTransferSyntax()
366 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_1_99);
370 * \brief Determines if the Transfer Syntax was already encountered
371 * and if it corresponds to a Explicit VR Big Endian one.
372 * @return True when big endian found. False in all other cases.
374 bool gdcmDocument::IsExplicitVRBigEndianTransferSyntax()
376 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_2);
380 * \brief Determines if the Transfer Syntax was already encountered
381 * and if it corresponds to a JPEGBaseLineProcess1 one.
382 * @return True when JPEGBaseLineProcess1found. False in all other cases.
384 bool gdcmDocument::IsJPEGBaseLineProcess1TransferSyntax()
386 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_50);
390 * \brief Determines if the Transfer Syntax was already encountered
391 * and if it corresponds to a JPEGExtendedProcess2-4 one.
392 * @return True when JPEGExtendedProcess2-4 found. False in all other cases.
394 bool gdcmDocument::IsJPEGExtendedProcess2_4TransferSyntax()
396 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_51);
400 * \brief Determines if the Transfer Syntax was already encountered
401 * and if it corresponds to a JPEGExtendeProcess3-5 one.
402 * @return True when JPEGExtendedProcess3-5 found. False in all other cases.
404 bool gdcmDocument::IsJPEGExtendedProcess3_5TransferSyntax()
406 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_52);
410 * \brief Determines if the Transfer Syntax was already encountered
411 * and if it corresponds to a JPEGSpectralSelectionProcess6-8 one.
412 * @return True when JPEGSpectralSelectionProcess6-8 found. False in all
415 bool gdcmDocument::IsJPEGSpectralSelectionProcess6_8TransferSyntax()
417 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_53);
421 * \brief Determines if the Transfer Syntax was already encountered
422 * and if it corresponds to a RLE Lossless one.
423 * @return True when RLE Lossless found. False in all
426 bool gdcmDocument::IsRLELossLessTransferSyntax()
428 return IsGivenTransferSyntax(UI1_2_840_10008_1_2_5);
432 * \brief Determines if Transfer Syntax was already encountered
433 * and if it corresponds to a JPEG Lossless one.
434 * @return True when RLE Lossless found. False in all
438 bool gdcmDocument::IsJPEGLossless()
440 return ( IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_55)
441 || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_57)
442 || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_70) );
446 * \brief Determines if the Transfer Syntax was already encountered
447 * and if it corresponds to a JPEG2000 one
448 * @return True when JPEG2000 (Lossly or LossLess) found. False in all
451 bool gdcmDocument::IsJPEG2000()
453 return ( IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_90)
454 || IsGivenTransferSyntax(UI1_2_840_10008_1_2_4_91) );
458 * \brief Predicate for dicom version 3 file.
459 * @return True when the file is a dicom version 3.
461 bool gdcmDocument::IsDicomV3()
463 // Checking if Transfert Syntax exists is enough
464 // Anyway, it's to late check if the 'Preamble' was found ...
465 // And ... would it be a rich idea to check ?
466 // (some 'no Preamble' DICOM images exist !)
467 return GetDocEntryByNumber(0x0002, 0x0010) != NULL;
471 * \brief returns the File Type
472 * (ACR, ACR_LIBIDO, ExplicitVR, ImplicitVR, Unknown)
473 * @return the FileType code
475 FileType gdcmDocument::GetFileType()
481 * \brief opens the file
482 * @param exception_on_error
485 FILE *gdcmDocument::OpenFile(bool exception_on_error)
488 Fp = fopen(Filename.c_str(),"rb");
492 if(exception_on_error)
494 throw gdcmFileError("gdcmDocument::gdcmDocument(const char *, bool)");
498 dbg.Verbose(0, "gdcmDocument::OpenFile cannot open file: ",
507 fread(&zero, (size_t)2, (size_t)1, Fp);
509 //ACR -- or DICOM with no Preamble --
510 if( zero == 0x0008 || zero == 0x0800 || zero == 0x0002 || zero == 0x0200 )
516 fseek(Fp, 126L, SEEK_CUR);
518 fread(dicm, (size_t)4, (size_t)1, Fp);
519 if( memcmp(dicm, "DICM", 4) == 0 )
525 dbg.Verbose(0, "gdcmDocument::OpenFile not DICOM/ACR", Filename.c_str());
529 dbg.Verbose(0, "gdcmDocument::OpenFile cannot open file", Filename.c_str());
536 * \brief closes the file
537 * @return TRUE if the close was successfull
539 bool gdcmDocument::CloseFile()
541 int closed = fclose(Fp);
548 * \brief Writes in a file all the Header Entries (Dicom Elements)
549 * @param fp file pointer on an already open file
550 * @param filetype Type of the File to be written
551 * (ACR-NEMA, ExplicitVR, ImplicitVR)
552 * \return Always true.
554 void gdcmDocument::Write(FILE* fp,FileType filetype)
556 /// \todo move the following lines (and a lot of others, to be written)
557 /// to a future function CheckAndCorrectHeader
559 /// WARNING : Si on veut ecrire du DICOM V3 a partir d'un DcmHeader ACR-NEMA
560 /// no way (check : FileType est un champ de gdcmDocument ...)
561 /// a moins de se livrer a un tres complique ajout des champs manquants.
562 /// faire un CheckAndCorrectHeader (?)
564 if (filetype == gdcmImplicitVR)
566 std::string implicitVRTransfertSyntax = UI1_2_840_10008_1_2;
567 ReplaceOrCreateByNumber(implicitVRTransfertSyntax,0x0002, 0x0010);
569 /// \todo Refer to standards on page 21, chapter 6.2
570 /// "Value representation": values with a VR of UI shall be
571 /// padded with a single trailing null
572 /// in the following case we have to padd manually with a 0
574 SetEntryLengthByNumber(18, 0x0002, 0x0010);
577 if (filetype == gdcmExplicitVR)
579 std::string explicitVRTransfertSyntax = UI1_2_840_10008_1_2_1;
580 ReplaceOrCreateByNumber(explicitVRTransfertSyntax,0x0002, 0x0010);
582 /// \todo Refer to standards on page 21, chapter 6.2
583 /// "Value representation": values with a VR of UI shall be
584 /// padded with a single trailing null
585 /// Dans le cas suivant on doit pader manuellement avec un 0
587 SetEntryLengthByNumber(20, 0x0002, 0x0010);
591 * \todo rewrite later, if really usefull
592 * - 'Group Length' element is optional in DICOM
593 * - but un-updated odd groups lengthes can causes pb
596 * if ( (filetype == ImplicitVR) || (filetype == ExplicitVR) )
597 * UpdateGroupLength(false,filetype);
598 * if ( filetype == ACR)
599 * UpdateGroupLength(true,ACR);
602 gdcmElementSet::Write(fp, filetype); // This one is recursive
607 * \brief Modifies the value of a given Header Entry (Dicom Element)
608 * when it exists. Create it with the given value when unexistant.
609 * @param Value (string) Value to be set
610 * @param Group Group number of the Entry
611 * @param Elem Element number of the Entry
612 * \return pointer to the modified/created Header Entry (NULL when creation
616 gdcmValEntry * gdcmDocument::ReplaceOrCreateByNumber(
621 gdcmValEntry* valEntry = 0;
623 gdcmDocEntry* currentEntry = GetDocEntryByNumber( group, elem);
626 // The entry wasn't present and we simply create the required ValEntry:
627 currentEntry = NewDocEntryByNumber(group, elem);
630 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: call to"
631 " NewDocEntryByNumber failed.");
634 valEntry = new gdcmValEntry(currentEntry);
635 if ( !AddEntry(valEntry))
637 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: AddEntry"
638 " failed allthough this is a creation.");
643 valEntry = dynamic_cast< gdcmValEntry* >(currentEntry);
646 // We need to promote the gdcmDocEntry to a gdcmValEntry:
647 valEntry = new gdcmValEntry(currentEntry);
648 if (!RemoveEntry(currentEntry))
650 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: removal"
651 " of previous DocEntry failed.");
654 if ( !AddEntry(valEntry))
656 dbg.Verbose(0, "gdcmDocument::ReplaceOrCreateByNumber: adding"
657 " promoted ValEntry failed.");
663 SetEntryByNumber(value, group, elem);
669 * \brief Modifies the value of a given Header Entry (Dicom Element)
670 * when it exists. Create it with the given value when unexistant.
671 * @param voidArea (binary) value to be set
672 * @param Group Group number of the Entry
673 * @param Elem Element number of the Entry
674 * \return pointer to the modified/created Header Entry (NULL when creation
677 gdcmBinEntry * gdcmDocument::ReplaceOrCreateByNumber(
684 gdcmDocEntry* a = GetDocEntryByNumber( group, elem);
687 a = NewBinEntryByNumber(group, elem);
693 b = new gdcmBinEntry(a);
695 b->SetVoidArea(voidArea);
697 SetEntryByNumber(voidArea, lgth, group, elem);
698 //b->SetVoidArea(voidArea); //what if b == 0 !!
704 * \brief Set a new value if the invoked element exists
705 * Seems to be useless !!!
706 * @param Value new element value
707 * @param Group group number of the Entry
708 * @param Elem element number of the Entry
711 bool gdcmDocument::ReplaceIfExistByNumber(const char* value, uint16_t group,
714 const std::string v = value;
715 SetEntryByNumber(v, group, elem);
720 //-----------------------------------------------------------------------------
724 * \brief Checks if a given Dicom Element exists within the H table
725 * @param group Group number of the searched Dicom Element
726 * @param element Element number of the searched Dicom Element
727 * @return number of occurences
729 int gdcmDocument::CheckIfEntryExistByNumber(uint16_t group, uint16_t element )
731 std::string key = gdcmDictEntry::TranslateToKey(group, element );
732 return TagHT.count(key);
736 * \brief Searches within Header Entries (Dicom Elements) parsed with
737 * the public and private dictionaries
738 * for the element value of a given tag.
739 * \warning Don't use any longer : use GetPubEntryByName
740 * @param tagName name of the searched element.
741 * @return Corresponding element value when it exists,
742 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
744 std::string gdcmDocument::GetEntryByName(TagName tagName)
746 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
752 return GetEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement());
756 * \brief Searches within Header Entries (Dicom Elements) parsed with
757 * the public and private dictionaries
758 * for the element value representation of a given tag.
760 * Obtaining the VR (Value Representation) might be needed by caller
761 * to convert the string typed content to caller's native type
762 * (think of C++ vs Python). The VR is actually of a higher level
763 * of semantics than just the native C++ type.
764 * @param tagName name of the searched element.
765 * @return Corresponding element value representation when it exists,
766 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
768 std::string gdcmDocument::GetEntryVRByName(TagName tagName)
770 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
771 if( dictEntry == NULL)
776 gdcmDocEntry* elem = GetDocEntryByNumber(dictEntry->GetGroup(),
777 dictEntry->GetElement());
778 return elem->GetVR();
783 * \brief Searches within Header Entries (Dicom Elements) parsed with
784 * the public and private dictionaries
785 * for the element value representation of a given tag.
786 * @param group Group number of the searched tag.
787 * @param element Element number of the searched tag.
788 * @return Corresponding element value representation when it exists,
789 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
791 std::string gdcmDocument::GetEntryByNumber(uint16_t group, uint16_t element)
793 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element);
794 /// \todo use map methods, instead of multimap JPR
795 if ( !TagHT.count(key))
800 return ((gdcmValEntry *)TagHT.find(key)->second)->GetValue();
804 * \brief Searches within Header Entries (Dicom Elements) parsed with
805 * the public and private dictionaries
806 * for the element value representation of a given tag..
808 * Obtaining the VR (Value Representation) might be needed by caller
809 * to convert the string typed content to caller's native type
810 * (think of C++ vs Python). The VR is actually of a higher level
811 * of semantics than just the native C++ type.
812 * @param group Group number of the searched tag.
813 * @param element Element number of the searched tag.
814 * @return Corresponding element value representation when it exists,
815 * and the string GDCM_UNFOUND ("gdcm::Unfound") otherwise.
817 std::string gdcmDocument::GetEntryVRByNumber(uint16_t group, uint16_t element)
819 gdcmDocEntry* elem = GetDocEntryByNumber(group, element);
824 return elem->GetVR();
828 * \brief Searches within Header Entries (Dicom Elements) parsed with
829 * the public and private dictionaries
830 * for the value length of a given tag..
831 * @param group Group number of the searched tag.
832 * @param element Element number of the searched tag.
833 * @return Corresponding element length; -2 if not found
835 int gdcmDocument::GetEntryLengthByNumber(uint16_t group, uint16_t element)
837 gdcmDocEntry* elem = GetDocEntryByNumber(group, element);
840 return -2; //magic number
842 return elem->GetLength();
845 * \brief Sets the value (string) of the Header Entry (Dicom Element)
846 * @param content string value of the Dicom Element
847 * @param tagName name of the searched Dicom Element.
848 * @return true when found
850 bool gdcmDocument::SetEntryByName(std::string content,std::string tagName)
852 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
858 return SetEntryByNumber(content,dictEntry->GetGroup(),
859 dictEntry->GetElement());
863 * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element)
864 * through it's (group, element) and modifies it's content with
866 * @param content new value (string) to substitute with
867 * @param group group number of the Dicom Element to modify
868 * @param element element number of the Dicom Element to modify
870 bool gdcmDocument::SetEntryByNumber(std::string content,
874 gdcmValEntry* valEntry = GetValEntryByNumber(group, element);
877 dbg.Verbose(0, "gdcmDocument::SetEntryByNumber: no corresponding",
878 " ValEntry (try promotion first).");
881 // Non even content must be padded with a space (020H).
882 if( content.length() % 2 )
886 valEntry->SetValue(content);
888 // Integers have a special treatement for their length:
889 gdcmVRKey vr = valEntry->GetVR();
890 if( vr == "US" || vr == "SS" )
892 valEntry->SetLength(2);
894 else if( vr == "UL" || vr == "SL" )
896 valEntry->SetLength(4);
900 valEntry->SetLength(content.length());
907 * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element)
908 * through it's (group, element) and modifies it's content with
910 * @param content new value (void *) to substitute with
911 * @param lgth new value length
912 * @param group group number of the Dicom Element to modify
913 * @param element element number of the Dicom Element to modify
915 bool gdcmDocument::SetEntryByNumber(void *content,
920 (void)lgth; //not used
921 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element);
922 if ( !TagHT.count(key))
927 /* Hope Binaray field length is *never* wrong
928 if(lgth%2) // Non even length are padded with a space (020H).
931 //content = content + '\0'; // fing a trick to enlarge a binary field?
935 a = (gdcmBinEntry *)TagHT[key];
936 a->SetVoidArea(content);
937 //a->SetLength(lgth); // ???
942 * \brief Accesses an existing gdcmDocEntry (i.e. a Dicom Element)
943 * in the PubDocEntrySet of this instance
944 * through it's (group, element) and modifies it's length with
946 * \warning Use with extreme caution.
947 * @param l new length to substitute with
948 * @param group group number of the Entry to modify
949 * @param element element number of the Entry to modify
950 * @return true on success, false otherwise.
952 bool gdcmDocument::SetEntryLengthByNumber(uint32_t l,
956 /// \todo use map methods, instead of multimap JPR
957 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element);
958 if ( !TagHT.count(key) )
964 l++; // length must be even
966 ( ((TagHT.equal_range(key)).first)->second )->SetLength(l);
972 * \brief Gets (from Header) the offset of a 'non string' element value
973 * (LoadElementValues has already be executed)
974 * @param Group group number of the Entry
975 * @param Elem element number of the Entry
976 * @return File Offset of the Element Value
978 size_t gdcmDocument::GetEntryOffsetByNumber(uint16_t group, uint16_t elem)
980 gdcmDocEntry* entry = GetDocEntryByNumber(group, elem);
983 dbg.Verbose(1, "gdcmDocument::GetDocEntryByNumber: no entry present.");
986 return entry->GetOffset();
990 * \brief Gets (from Header) a 'non string' element value
991 * (LoadElementValues has already be executed)
992 * @param Group group number of the Entry
993 * @param Elem element number of the Entry
994 * @return Pointer to the 'non string' area
996 void * gdcmDocument::GetEntryVoidAreaByNumber(uint16_t group, uint16_t elem)
998 gdcmDocEntry* entry = GetDocEntryByNumber(group, elem);
1001 dbg.Verbose(1, "gdcmDocument::GetDocEntryByNumber: no entry");
1004 return ((gdcmBinEntry *)entry)->GetVoidArea();
1008 * \brief Loads (from disk) the element content
1009 * when a string is not suitable
1010 * @param Group group number of the Entry
1011 * @param Elem element number of the Entry
1013 void *gdcmDocument::LoadEntryVoidArea(uint16_t group, uint16_t elem)
1015 gdcmDocEntry *docElement = GetDocEntryByNumber(group, elem);
1020 size_t o =(size_t)docElement->GetOffset();
1021 fseek(Fp, o, SEEK_SET);
1022 size_t l = docElement->GetLength();
1023 char* a = new char[l];
1026 dbg.Verbose(0, "gdcmDocument::LoadEntryVoidArea cannot allocate a");
1029 size_t l2 = fread(a, 1, l , Fp);
1035 /// \todo Drop any already existing void area! JPR
1036 SetEntryVoidAreaByNumber(a, group, elem);
1041 * \brief Loads (from disk) the element content
1042 * when a string is not suitable
1043 * @param Element Entry whose voidArea is going to be loaded
1045 void *gdcmDocument::LoadEntryVoidArea(gdcmBinEntry *element)
1047 size_t o =(size_t)element->GetOffset();
1048 fseek(Fp, o, SEEK_SET);
1049 size_t l = element->GetLength();
1050 char* a = new char[l];
1053 dbg.Verbose(0, "gdcmDocument::LoadEntryVoidArea cannot allocate a");
1056 element->SetVoidArea((void *)a);
1057 /// \todo check the result
1058 size_t l2 = fread(a, 1, l , Fp);
1069 * \brief Sets a 'non string' value to a given Dicom Element
1070 * @param area area containing the 'non string' value
1071 * @param group Group number of the searched Dicom Element
1072 * @param element Element number of the searched Dicom Element
1075 bool gdcmDocument::SetEntryVoidAreaByNumber(void * area,
1079 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element);
1080 if ( !TagHT.count(key))
1084 // This was for multimap ?
1085 (( gdcmBinEntry *)( ((TagHT.equal_range(key)).first)->second ))->SetVoidArea(area);
1091 * \brief Update the entries with the shadow dictionary.
1092 * Only non even entries are analyzed
1094 void gdcmDocument::UpdateShaEntries()
1096 //gdcmDictEntry *entry;
1099 /// \todo TODO : still any use to explore recursively the whole structure?
1101 for(ListTag::iterator it=listEntries.begin();
1102 it!=listEntries.end();
1105 // Odd group => from public dictionary
1106 if((*it)->GetGroup()%2==0)
1109 // Peer group => search the corresponding dict entry
1111 entry=RefShaDict->GetDictEntryByNumber((*it)->GetGroup(),(*it)->GetElement());
1115 if((*it)->IsImplicitVR())
1120 (*it)->SetValue(GetDocEntryUnvalue(*it)); // to go on compiling
1122 // Set the new entry and the new value
1123 (*it)->SetDictEntry(entry);
1124 CheckDocEntryVR(*it,vr);
1126 (*it)->SetValue(GetDocEntryValue(*it)); // to go on compiling
1131 // Remove precedent value transformation
1132 (*it)->SetDictEntry(NewVirtualDictEntry((*it)->GetGroup(),(*it)->GetElement(),vr));
1139 * \brief Searches within the Header Entries for a Dicom Element of
1141 * @param tagName name of the searched Dicom Element.
1142 * @return Corresponding Dicom Element when it exists, and NULL
1145 gdcmDocEntry* gdcmDocument::GetDocEntryByName(std::string const & tagName)
1147 gdcmDictEntry *dictEntry = RefPubDict->GetDictEntryByName(tagName);
1153 return GetDocEntryByNumber(dictEntry->GetGroup(),dictEntry->GetElement());
1157 * \brief retrieves a Dicom Element (the first one) using (group, element)
1158 * \warning (group, element) IS NOT an identifier inside the Dicom Header
1159 * if you think it's NOT UNIQUE, check the count number
1160 * and use iterators to retrieve ALL the Dicoms Elements within
1161 * a given couple (group, element)
1162 * @param group Group number of the searched Dicom Element
1163 * @param element Element number of the searched Dicom Element
1166 gdcmDocEntry* gdcmDocument::GetDocEntryByNumber(uint16_t group,
1169 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, element);
1170 if ( !TagHT.count(key))
1174 return TagHT.find(key)->second;
1178 * \brief Same as \ref gdcmDocument::GetDocEntryByNumber except it only
1179 * returns a result when the corresponding entry is of type
1181 * @return When present, the corresponding ValEntry.
1183 gdcmValEntry* gdcmDocument::GetValEntryByNumber(uint16_t group,
1186 gdcmDocEntry* currentEntry = GetDocEntryByNumber(group, element);
1187 if ( !currentEntry )
1191 if ( gdcmValEntry* valEntry = dynamic_cast<gdcmValEntry*>(currentEntry) )
1195 dbg.Verbose(0, "gdcmDocument::GetValEntryByNumber: unfound ValEntry.");
1201 * \brief Loads the element while preserving the current
1202 * underlying file position indicator as opposed to
1203 * to LoadDocEntry that modifies it.
1204 * @param entry Header Entry whose value shall be loaded.
1207 void gdcmDocument::LoadDocEntrySafe(gdcmDocEntry * entry)
1209 long PositionOnEntry = ftell(Fp);
1210 LoadDocEntry(entry);
1211 fseek(Fp, PositionOnEntry, SEEK_SET);
1215 * \brief Swaps back the bytes of 4-byte long integer accordingly to
1217 * @return The properly swaped 32 bits integer.
1219 uint32_t gdcmDocument::SwapLong(uint32_t a)
1226 a=( ((a<<24) & 0xff000000) | ((a<<8) & 0x00ff0000) |
1227 ((a>>8) & 0x0000ff00) | ((a>>24) & 0x000000ff) );
1231 a=( ((a<<16) & 0xffff0000) | ((a>>16) & 0x0000ffff) );
1235 a=( ((a<< 8) & 0xff00ff00) | ((a>>8) & 0x00ff00ff) );
1238 std::cout << "swapCode= " << SwapCode << std::endl;
1239 dbg.Error(" gdcmDocument::SwapLong : unset swap code");
1246 * \brief Unswaps back the bytes of 4-byte long integer accordingly to
1248 * @return The properly unswaped 32 bits integer.
1250 uint32_t gdcmDocument::UnswapLong(uint32_t a)
1256 * \brief Swaps the bytes so they agree with the processor order
1257 * @return The properly swaped 16 bits integer.
1259 uint16_t gdcmDocument::SwapShort(uint16_t a)
1261 if ( SwapCode == 4321 || SwapCode == 2143 )
1263 a =((( a << 8 ) & 0x0ff00 ) | (( a >> 8 ) & 0x00ff ) );
1269 * \brief Unswaps the bytes so they agree with the processor order
1270 * @return The properly unswaped 16 bits integer.
1272 uint16_t gdcmDocument::UnswapShort(uint16_t a)
1274 return SwapShort(a);
1277 //-----------------------------------------------------------------------------
1281 * \brief Parses a DocEntrySet (Zero-level DocEntries or SQ Item DocEntries)
1282 * @return length of the parsed set.
1285 long gdcmDocument::ParseDES(gdcmDocEntrySet *set,
1290 gdcmDocEntry *newDocEntry = 0;
1291 gdcmValEntry *newValEntry = 0;
1292 unsigned long l = 0;
1294 int depth = set->GetDepthLevel();
1297 if ( !delim_mode && (ftell(Fp)-offset) >= l_max)
1301 newDocEntry = ReadNextDocEntry( );
1307 gdcmVRKey vr = newDocEntry->GetVR();
1311 if ( gdcmGlobal::GetVR()->IsVROfGdcmStringRepresentable(vr) )
1314 newValEntry = new gdcmValEntry(newDocEntry->GetDictEntry());
1315 newValEntry->Copy(newDocEntry);
1316 newValEntry->SetDepthLevel(depth);
1317 set->AddEntry(newValEntry);
1318 LoadDocEntry(newValEntry);
1319 if (newValEntry->isItemDelimitor())
1323 if ( !delim_mode && (ftell(Fp)-offset) >= l_max)
1330 if ( ! gdcmGlobal::GetVR()->IsVROfGdcmBinaryRepresentable(vr) )
1332 ////// Neither ValEntry NOR BinEntry: should mean UNKOWN VR
1333 dbg.Verbose(0, "gdcmDocument::ParseDES: neither Valentry, "
1334 "nor BinEntry. Probably unknown VR.");
1337 ////// BinEntry or UNKOWN VR:
1338 gdcmBinEntry *bn = new gdcmBinEntry(newDocEntry->GetDictEntry());
1339 bn->Copy(newDocEntry);
1344 if (newDocEntry->GetGroup() == 0x7fe0 &&
1345 newDocEntry->GetElement() == 0x0010 )
1347 if (newDocEntry->GetReadLength()==0xffffffff)
1349 // Broken US.3405.1.dcm
1350 Parse7FE0(); // to skip the pixels
1351 // (multipart JPEG/RLE are trouble makers)
1355 SkipToNextDocEntry(newDocEntry);
1356 l = newDocEntry->GetFullLength();
1361 // to be sure we are at the beginning
1362 SkipToNextDocEntry(newDocEntry);
1363 l = newDocEntry->GetFullLength();
1369 l = newDocEntry->GetReadLength();
1370 if ( l != 0 ) // don't mess the delim_mode for zero-length sequence
1372 if ( l == 0xffffffff )
1381 // no other way to create it ...
1382 gdcmSeqEntry *sq = new gdcmSeqEntry(newDocEntry->GetDictEntry(),
1383 set->GetDepthLevel());
1384 sq->Copy(newDocEntry);
1385 sq->SetDelimitorMode(delim_mode);
1386 sq->SetDepthLevel(depth);
1389 { // Don't try to parse zero-length sequences
1390 long lgt = ParseSQ( sq,
1391 newDocEntry->GetOffset(),
1393 (void)lgt; //not used...
1396 if ( !delim_mode && (ftell(Fp)-offset) >= l_max)
1404 return l; // Probably useless
1408 * \brief Parses a Sequence ( SeqEntry after SeqEntry)
1409 * @return parsed length for this level
1411 long gdcmDocument::ParseSQ(gdcmSeqEntry *set,
1412 long offset, long l_max, bool delim_mode)
1414 int SQItemNumber = 0;
1416 //int depth = set->GetDepthLevel();
1417 //(void)depth; //not used
1421 gdcmDocEntry *newDocEntry = ReadNextDocEntry();
1424 // FIXME Should warn user
1429 if ( newDocEntry->isSequenceDelimitor() )
1431 set->SetSequenceDelimitationItem( newDocEntry );
1435 if ( !delim_mode && (ftell(Fp)-offset) >= l_max)
1440 gdcmSQItem *itemSQ = new gdcmSQItem(set->GetDepthLevel());
1441 itemSQ->AddEntry(newDocEntry);
1442 unsigned int l = newDocEntry->GetReadLength();
1444 if ( l == 0xffffffff )
1453 int lgr = ParseDES(itemSQ, newDocEntry->GetOffset(), l, dlm_mod);
1455 set->AddEntry(itemSQ, SQItemNumber);
1457 if ( !delim_mode && (ftell(Fp)-offset) >= l_max)
1463 int lgth = ftell(Fp) - offset;
1468 * \brief Loads the element content if its length doesn't exceed
1469 * the value specified with gdcmDocument::SetMaxSizeLoadEntry()
1470 * @param Entry Header Entry (Dicom Element) to be dealt with
1472 void gdcmDocument::LoadDocEntry(gdcmDocEntry* entry)
1475 uint16_t group = entry->GetGroup();
1476 std::string vr = entry->GetVR();
1477 uint32_t length = entry->GetLength();
1479 fseek(Fp, (long)entry->GetOffset(), SEEK_SET);
1481 // A SeQuence "contains" a set of Elements.
1482 // (fffe e000) tells us an Element is beginning
1483 // (fffe e00d) tells us an Element just ended
1484 // (fffe e0dd) tells us the current SeQuence just ended
1485 if( group == 0xfffe )
1487 // NO more value field for SQ !
1491 // When the length is zero things are easy:
1494 ((gdcmValEntry *)entry)->SetValue("");
1498 // The elements whose length is bigger than the specified upper bound
1499 // are not loaded. Instead we leave a short notice of the offset of
1500 // the element content and it's length.
1502 std::ostringstream s;
1503 if (length > MaxSizeLoadEntry)
1505 if (gdcmBinEntry* binEntryPtr = dynamic_cast< gdcmBinEntry* >(entry) )
1507 s << "gdcm::NotLoaded (BinEntry)";
1508 s << " Address:" << (long)entry->GetOffset();
1509 s << " Length:" << entry->GetLength();
1510 s << " x(" << std::hex << entry->GetLength() << ")";
1511 binEntryPtr->SetValue(s.str());
1513 // to be sure we are at the end of the value ...
1514 fseek(Fp, (long)entry->GetOffset()+(long)entry->GetLength(), SEEK_SET);
1516 return; //FIXME FIXME FIXME FIXME ????
1518 // Be carefull : a BinEntry IS_A ValEntry ...
1519 if (gdcmValEntry* valEntryPtr = dynamic_cast< gdcmValEntry* >(entry) )
1521 s << "gdcm::NotLoaded. (ValEntry)";
1522 s << " Address:" << (long)entry->GetOffset();
1523 s << " Length:" << entry->GetLength();
1524 s << " x(" << std::hex << entry->GetLength() << ")";
1525 valEntryPtr->SetValue(s.str());
1527 // to be sure we are at the end of the value ...
1528 fseek(Fp,(long)entry->GetOffset()+(long)entry->GetLength(),SEEK_SET);
1533 // When we find a BinEntry not very much can be done :
1534 if (gdcmBinEntry* binEntryPtr = dynamic_cast< gdcmBinEntry* >(entry) )
1537 LoadEntryVoidArea(binEntryPtr);
1538 s << "gdcm::Loaded (BinEntry)";
1539 binEntryPtr->SetValue(s.str());
1543 /// \todo Any compacter code suggested (?)
1544 if ( IsDocEntryAnInteger(entry) )
1547 std::ostringstream s;
1549 // When short integer(s) are expected, read and convert the following
1550 // n *two characters properly i.e. consider them as short integers as
1551 // opposed to strings.
1552 // Elements with Value Multiplicity > 1
1553 // contain a set of integers (not a single one)
1554 if (vr == "US" || vr == "SS")
1557 NewInt = ReadInt16();
1561 for (int i=1; i < nbInt; i++)
1564 NewInt = ReadInt16();
1569 // See above comment on multiple integers (mutatis mutandis).
1570 else if (vr == "UL" || vr == "SL")
1573 NewInt = ReadInt32();
1577 for (int i=1; i < nbInt; i++)
1580 NewInt = ReadInt32();
1585 #ifdef GDCM_NO_ANSI_STRING_STREAM
1586 s << std::ends; // to avoid oddities on Solaris
1587 #endif //GDCM_NO_ANSI_STRING_STREAM
1589 ((gdcmValEntry *)entry)->SetValue(s.str());
1593 // We need an additional byte for storing \0 that is not on disk
1594 //std::string newValue(length,0);
1595 //item_read = fread(&(newValue[0]), (size_t)length, (size_t)1, Fp);
1596 //rah !! I can't believe it could work, normally this is a const char* !!!
1597 char *str = new char[length+1];
1598 item_read = fread(str, (size_t)length, (size_t)1, Fp);
1600 std::string newValue = str;
1602 if ( gdcmValEntry* valEntry = dynamic_cast<gdcmValEntry* >(entry) )
1604 if ( item_read != 1 )
1606 dbg.Verbose(1, "gdcmDocument::LoadDocEntry",
1607 "unread element value");
1608 valEntry->SetValue("gdcm::UnRead");
1614 // Because of correspondance with the VR dic
1615 valEntry->SetValue(newValue);
1619 valEntry->SetValue(newValue);
1624 dbg.Error(true, "gdcmDocument::LoadDocEntry"
1625 "Should have a ValEntry, here !");
1631 * \brief Find the value Length of the passed Header Entry
1632 * @param Entry Header Entry whose length of the value shall be loaded.
1634 void gdcmDocument::FindDocEntryLength (gdcmDocEntry *entry)
1636 uint16_t element = entry->GetElement();
1637 std::string vr = entry->GetVR();
1641 if ( Filetype == gdcmExplicitVR && !entry->IsImplicitVR() )
1643 if ( vr == "OB" || vr == "OW" || vr == "SQ" || vr == "UN" )
1645 // The following reserved two bytes (see PS 3.5-2001, section
1646 // 7.1.2 Data element structure with explicit vr p27) must be
1647 // skipped before proceeding on reading the length on 4 bytes.
1648 fseek(Fp, 2L, SEEK_CUR);
1649 uint32_t length32 = ReadInt32();
1651 if ( vr == "OB" && length32 == 0xffffffff )
1653 uint32_t lengthOB = FindDocEntryLengthOB();
1656 // Computing the length failed (this happens with broken
1657 // files like gdcm-JPEG-LossLess3a.dcm). We still have a
1658 // chance to get the pixels by deciding the element goes
1659 // until the end of the file. Hence we artificially fix the
1660 // the length and proceed.
1661 long currentPosition = ftell(Fp);
1662 fseek(Fp,0L,SEEK_END);
1663 long lengthUntilEOF = ftell(Fp) - currentPosition;
1664 fseek(Fp, currentPosition, SEEK_SET);
1665 entry->SetLength(lengthUntilEOF);
1669 entry->SetLength(lengthOB);
1672 FixDocEntryFoundLength(entry, length32);
1676 // Length is encoded on 2 bytes.
1677 length16 = ReadInt16();
1679 // We can tell the current file is encoded in big endian (like
1680 // Data/US-RGB-8-epicard) when we find the "Transfer Syntax" tag
1681 // and it's value is the one of the encoding of a big endian file.
1682 // In order to deal with such big endian encoded files, we have
1683 // (at least) two strategies:
1684 // * when we load the "Transfer Syntax" tag with value of big endian
1685 // encoding, we raise the proper flags. Then we wait for the end
1686 // of the META group (0x0002) among which is "Transfer Syntax",
1687 // before switching the swap code to big endian. We have to postpone
1688 // the switching of the swap code since the META group is fully encoded
1689 // in little endian, and big endian coding only starts at the next
1690 // group. The corresponding code can be hard to analyse and adds
1691 // many additional unnecessary tests for regular tags.
1692 // * the second strategy consists in waiting for trouble, that shall
1693 // appear when we find the first group with big endian encoding. This
1694 // is easy to detect since the length of a "Group Length" tag (the
1695 // ones with zero as element number) has to be of 4 (0x0004). When we
1696 // encounter 1024 (0x0400) chances are the encoding changed and we
1697 // found a group with big endian encoding.
1698 // We shall use this second strategy. In order to make sure that we
1699 // can interpret the presence of an apparently big endian encoded
1700 // length of a "Group Length" without committing a big mistake, we
1701 // add an additional check: we look in the already parsed elements
1702 // for the presence of a "Transfer Syntax" whose value has to be "big
1703 // endian encoding". When this is the case, chances are we have got our
1704 // hands on a big endian encoded file: we switch the swap code to
1705 // big endian and proceed...
1706 if ( (element == 0x0000) && (length16 == 0x0400) )
1708 if ( ! IsExplicitVRBigEndianTransferSyntax() )
1710 dbg.Verbose(0, "gdcmDocument::FindLength", "not explicit VR");
1715 SwitchSwapToBigEndian();
1716 // Restore the unproperly loaded values i.e. the group, the element
1717 // and the dictionary entry depending on them.
1718 uint16_t correctGroup = SwapShort( entry->GetGroup() );
1719 uint16_t correctElem = SwapShort( entry->GetElement() );
1720 gdcmDictEntry* newTag = GetDictEntryByNumber( correctGroup,
1724 // This correct tag is not in the dictionary. Create a new one.
1725 newTag = NewVirtualDictEntry(correctGroup, correctElem);
1727 // FIXME this can create a memory leaks on the old entry that be
1728 // left unreferenced.
1729 entry->SetDictEntry( newTag );
1732 // Heuristic: well, some files are really ill-formed.
1733 if ( length16 == 0xffff)
1736 // Length16= 0xffff means that we deal with
1737 // 'Unknown Length' Sequence
1739 FixDocEntryFoundLength(entry, (uint32_t)length16);
1744 // Either implicit VR or a non DICOM conformal (see note below) explicit
1745 // VR that ommited the VR of (at least) this element. Farts happen.
1746 // [Note: according to the part 5, PS 3.5-2001, section 7.1 p25
1747 // on Data elements "Implicit and Explicit VR Data Elements shall
1748 // not coexist in a Data Set and Data Sets nested within it".]
1749 // Length is on 4 bytes.
1751 FixDocEntryFoundLength(entry, ReadInt32());
1757 * \brief Find the Value Representation of the current Dicom Element.
1760 void gdcmDocument::FindDocEntryVR( gdcmDocEntry *entry )
1762 if ( Filetype != gdcmExplicitVR )
1769 long positionOnEntry = ftell(Fp);
1770 // Warning: we believe this is explicit VR (Value Representation) because
1771 // we used a heuristic that found "UL" in the first tag. Alas this
1772 // doesn't guarantee that all the tags will be in explicit VR. In some
1773 // cases (see e-film filtered files) one finds implicit VR tags mixed
1774 // within an explicit VR file. Hence we make sure the present tag
1775 // is in explicit VR and try to fix things if it happens not to be
1778 fread (&vr, (size_t)2,(size_t)1, Fp);
1781 if( !CheckDocEntryVR(entry, vr) )
1783 fseek(Fp, positionOnEntry, SEEK_SET);
1784 // When this element is known in the dictionary we shall use, e.g. for
1785 // the semantics (see the usage of IsAnInteger), the VR proposed by the
1786 // dictionary entry. Still we have to flag the element as implicit since
1787 // we know now our assumption on expliciteness is not furfilled.
1789 if ( entry->IsVRUnknown() )
1791 entry->SetVR("Implicit");
1793 entry->SetImplicitVR();
1798 * \brief Check the correspondance between the VR of the header entry
1799 * and the taken VR. If they are different, the header entry is
1800 * updated with the new VR.
1801 * @param Entry Header Entry to check
1802 * @param vr Dicom Value Representation
1803 * @return false if the VR is incorrect of if the VR isn't referenced
1804 * otherwise, it returns true
1806 bool gdcmDocument::CheckDocEntryVR(gdcmDocEntry *entry, gdcmVRKey vr)
1808 char msg[100]; // for sprintf
1809 bool realExplicit = true;
1811 // Assume we are reading a falsely explicit VR file i.e. we reached
1812 // a tag where we expect reading a VR but are in fact we read the
1813 // first to bytes of the length. Then we will interogate (through find)
1814 // the dicom_vr dictionary with oddities like "\004\0" which crashes
1815 // both GCC and VC++ implementations of the STL map. Hence when the
1816 // expected VR read happens to be non-ascii characters we consider
1817 // we hit falsely explicit VR tag.
1819 if ( (!isalpha(vr[0])) && (!isalpha(vr[1])) )
1821 realExplicit = false;
1824 // CLEANME searching the dicom_vr at each occurence is expensive.
1825 // PostPone this test in an optional integrity check at the end
1826 // of parsing or only in debug mode.
1827 if ( realExplicit && !gdcmGlobal::GetVR()->Count(vr) )
1829 realExplicit = false;
1832 if ( !realExplicit )
1834 // We thought this was explicit VR, but we end up with an
1835 // implicit VR tag. Let's backtrack.
1836 sprintf(msg,"Falsely explicit vr file (%04x,%04x)\n",
1837 entry->GetGroup(), entry->GetElement());
1838 dbg.Verbose(1, "gdcmDocument::FindVR: ",msg);
1840 if( entry->GetGroup() % 2 && entry->GetElement() == 0x0000)
1842 // Group length is UL !
1843 gdcmDictEntry* newEntry = NewVirtualDictEntry(
1844 entry->GetGroup(), entry->GetElement(),
1845 "UL", "FIXME", "Group Length");
1846 entry->SetDictEntry( newEntry );
1851 if ( entry->IsVRUnknown() )
1853 // When not a dictionary entry, we can safely overwrite the VR.
1854 if( entry->GetElement() == 0x0000 )
1856 // Group length is UL !
1864 else if ( entry->GetVR() != vr )
1866 // The VR present in the file and the dictionary disagree. We assume
1867 // the file writer knew best and use the VR of the file. Since it would
1868 // be unwise to overwrite the VR of a dictionary (since it would
1869 // compromise it's next user), we need to clone the actual DictEntry
1870 // and change the VR for the read one.
1871 gdcmDictEntry* newEntry = NewVirtualDictEntry(
1872 entry->GetGroup(), entry->GetElement(),
1873 vr, "FIXME", entry->GetName());
1874 entry->SetDictEntry(newEntry);
1881 * \brief Get the transformed value of the header entry. The VR value
1882 * is used to define the transformation to operate on the value
1883 * \warning NOT end user intended method !
1885 * @return Transformed entry value
1887 std::string gdcmDocument::GetDocEntryValue(gdcmDocEntry *entry)
1889 if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() )
1891 std::string val = ((gdcmValEntry *)entry)->GetValue();
1892 std::string vr = entry->GetVR();
1893 uint32_t length = entry->GetLength();
1894 std::ostringstream s;
1897 // When short integer(s) are expected, read and convert the following
1898 // n * 2 bytes properly i.e. as a multivaluated strings
1899 // (each single value is separated fromthe next one by '\'
1900 // as usual for standard multivaluated filels
1901 // Elements with Value Multiplicity > 1
1902 // contain a set of short integers (not a single one)
1904 if( vr == "US" || vr == "SS" )
1909 for (int i=0; i < nbInt; i++)
1915 newInt16 = ( val[2*i+0] & 0xFF ) + ( ( val[2*i+1] & 0xFF ) << 8);
1916 newInt16 = SwapShort( newInt16 );
1921 // When integer(s) are expected, read and convert the following
1922 // n * 4 bytes properly i.e. as a multivaluated strings
1923 // (each single value is separated fromthe next one by '\'
1924 // as usual for standard multivaluated filels
1925 // Elements with Value Multiplicity > 1
1926 // contain a set of integers (not a single one)
1927 else if( vr == "UL" || vr == "SL" )
1932 for (int i=0; i < nbInt; i++)
1938 newInt32 = ( val[4*i+0] & 0xFF )
1939 + (( val[4*i+1] & 0xFF ) << 8 )
1940 + (( val[4*i+2] & 0xFF ) << 16 )
1941 + (( val[4*i+3] & 0xFF ) << 24 );
1942 newInt32 = SwapLong( newInt32 );
1946 #ifdef GDCM_NO_ANSI_STRING_STREAM
1947 s << std::ends; // to avoid oddities on Solaris
1948 #endif //GDCM_NO_ANSI_STRING_STREAM
1952 return ((gdcmValEntry *)entry)->GetValue();
1956 * \brief Get the reverse transformed value of the header entry. The VR
1957 * value is used to define the reverse transformation to operate on
1959 * \warning NOT end user intended method !
1961 * @return Reverse transformed entry value
1963 std::string gdcmDocument::GetDocEntryUnvalue(gdcmDocEntry* entry)
1965 if ( IsDocEntryAnInteger(entry) && entry->IsImplicitVR() )
1967 std::string vr = entry->GetVR();
1968 std::vector<std::string> tokens;
1969 std::ostringstream s;
1971 if ( vr == "US" || vr == "SS" )
1975 tokens.erase( tokens.begin(), tokens.end()); // clean any previous value
1976 Tokenize (((gdcmValEntry *)entry)->GetValue(), tokens, "\\");
1977 for (unsigned int i=0; i<tokens.size(); i++)
1979 newInt16 = atoi(tokens[i].c_str());
1980 s << ( newInt16 & 0xFF )
1981 << (( newInt16 >> 8 ) & 0xFF );
1985 if ( vr == "UL" || vr == "SL")
1989 tokens.erase(tokens.begin(),tokens.end()); // clean any previous value
1990 Tokenize (((gdcmValEntry *)entry)->GetValue(), tokens, "\\");
1991 for (unsigned int i=0; i<tokens.size();i++)
1993 newInt32 = atoi(tokens[i].c_str());
1994 s << (char)( newInt32 & 0xFF )
1995 << (char)(( newInt32 >> 8 ) & 0xFF )
1996 << (char)(( newInt32 >> 16 ) & 0xFF )
1997 << (char)(( newInt32 >> 24 ) & 0xFF );
2002 #ifdef GDCM_NO_ANSI_STRING_STREAM
2003 s << std::ends; // to avoid oddities on Solaris
2004 #endif //GDCM_NO_ANSI_STRING_STREAM
2008 return ((gdcmValEntry *)entry)->GetValue();
2012 * \brief Skip a given Header Entry
2013 * \warning NOT end user intended method !
2016 void gdcmDocument::SkipDocEntry(gdcmDocEntry *entry)
2018 SkipBytes(entry->GetLength());
2022 * \brief Skips to the begining of the next Header Entry
2023 * \warning NOT end user intended method !
2026 void gdcmDocument::SkipToNextDocEntry(gdcmDocEntry *entry)
2028 fseek(Fp, (long)(entry->GetOffset()), SEEK_SET);
2029 fseek(Fp, (long)(entry->GetReadLength()), SEEK_CUR);
2033 * \brief When the length of an element value is obviously wrong (because
2034 * the parser went Jabberwocky) one can hope improving things by
2035 * applying some heuristics.
2037 void gdcmDocument::FixDocEntryFoundLength(gdcmDocEntry *entry,
2038 uint32_t foundLength)
2040 entry->SetReadLength( foundLength ); // will be updated only if a bug is found
2041 if ( foundLength == 0xffffffff)
2046 uint16_t gr = entry->GetGroup();
2047 uint16_t el = entry->GetElement();
2049 if ( foundLength % 2)
2051 std::ostringstream s;
2052 s << "Warning : Tag with uneven length "
2054 << " in x(" << std::hex << gr << "," << el <<")" << std::dec;
2055 dbg.Verbose(0, s.str().c_str());
2058 //////// Fix for some naughty General Electric images.
2059 // Allthough not recent many such GE corrupted images are still present
2060 // on Creatis hard disks. Hence this fix shall remain when such images
2061 // are no longer in user (we are talking a few years, here)...
2062 // Note: XMedCom probably uses such a trick since it is able to read
2063 // those pesky GE images ...
2064 if ( foundLength == 13)
2066 // Only happens for this length !
2067 if ( entry->GetGroup() != 0x0008
2068 || ( entry->GetElement() != 0x0070
2069 && entry->GetElement() != 0x0080 ) )
2072 entry->SetReadLength(10); /// \todo a bug is to be fixed !?
2076 //////// Fix for some brain-dead 'Leonardo' Siemens images.
2077 // Occurence of such images is quite low (unless one leaves close to a
2078 // 'Leonardo' source. Hence, one might consider commenting out the
2079 // following fix on efficiency reasons.
2080 else if ( entry->GetGroup() == 0x0009
2081 && ( entry->GetElement() == 0x1113
2082 || entry->GetElement() == 0x1114 ) )
2085 entry->SetReadLength(4); /// \todo a bug is to be fixed !?
2088 //////// Deal with sequences, but only on users request:
2089 else if ( entry->GetVR() == "SQ" && EnableSequences)
2091 foundLength = 0; // ReadLength is unchanged
2094 //////// We encountered a 'delimiter' element i.e. a tag of the form
2095 // "fffe|xxxx" which is just a marker. Delimiters length should not be
2096 // taken into account.
2097 else if( entry->GetGroup() == 0xfffe )
2099 // According to the norm, fffe|0000 shouldn't exist. BUT the Philips
2100 // image gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm happens to
2101 // causes extra troubles...
2102 if( entry->GetElement() != 0x0000 )
2108 entry->SetUsableLength(foundLength);
2112 * \brief Apply some heuristics to predict whether the considered
2113 * element value contains/represents an integer or not.
2114 * @param Entry The element value on which to apply the predicate.
2115 * @return The result of the heuristical predicate.
2117 bool gdcmDocument::IsDocEntryAnInteger(gdcmDocEntry *entry)
2119 uint16_t element = entry->GetElement();
2120 uint16_t group = entry->GetGroup();
2121 std::string vr = entry->GetVR();
2122 uint32_t length = entry->GetLength();
2124 // When we have some semantics on the element we just read, and if we
2125 // a priori know we are dealing with an integer, then we shall be
2126 // able to swap it's element value properly.
2127 if ( element == 0 ) // This is the group length of the group
2135 // Allthough this should never happen, still some images have a
2136 // corrupted group length [e.g. have a glance at offset x(8336) of
2137 // gdcmData/gdcm-MR-PHILIPS-16-Multi-Seq.dcm].
2138 // Since for dicom compliant and well behaved headers, the present
2139 // test is useless (and might even look a bit paranoid), when we
2140 // encounter such an ill-formed image, we simply display a warning
2141 // message and proceed on parsing (while crossing fingers).
2142 std::ostringstream s;
2143 int filePosition = ftell(Fp);
2144 s << "Erroneous Group Length element length on : (" \
2145 << std::hex << group << " , " << element
2146 << ") -before- position x(" << filePosition << ")"
2147 << "lgt : " << length;
2148 dbg.Verbose(0, "gdcmDocument::IsDocEntryAnInteger", s.str().c_str() );
2152 if ( vr == "UL" || vr == "US" || vr == "SL" || vr == "SS" )
2161 * \brief Find the Length till the next sequence delimiter
2162 * \warning NOT end user intended method !
2166 uint32_t gdcmDocument::FindDocEntryLengthOB()
2168 // See PS 3.5-2001, section A.4 p. 49 on encapsulation of encoded pixel data.
2169 long positionOnEntry = ftell(Fp);
2170 bool foundSequenceDelimiter = false;
2171 uint32_t totalLength = 0;
2173 while ( !foundSequenceDelimiter )
2175 uint16_t g = ReadInt16();
2176 uint16_t n = ReadInt16();
2182 // We have to decount the group and element we just read
2185 if ( g != 0xfffe || ( n != 0xe0dd && n != 0xe000 ) )
2187 dbg.Verbose(1, "gdcmDocument::FindLengthOB: neither an Item tag "
2188 "nor a Sequence delimiter tag.");
2189 fseek(Fp, positionOnEntry, SEEK_SET);
2196 foundSequenceDelimiter = true;
2199 uint32_t itemLength = ReadInt32();
2200 // We add 4 bytes since we just read the ItemLength with ReadInt32
2201 totalLength += itemLength + 4;
2202 SkipBytes(itemLength);
2204 if ( foundSequenceDelimiter )
2209 fseek(Fp, positionOnEntry, SEEK_SET);
2214 * \brief Reads a supposed to be 16 Bits integer
2215 * (swaps it depending on processor endianity)
2216 * @return read value
2218 uint16_t gdcmDocument::ReadInt16()
2221 size_t item_read = fread (&g, (size_t)2,(size_t)1, Fp);
2222 if ( item_read != 1 )
2226 dbg.Verbose(0, "gdcmDocument::ReadInt16", " File Error");
2237 * \brief Reads a supposed to be 32 Bits integer
2238 * (swaps it depending on processor endianity)
2239 * @return read value
2241 uint32_t gdcmDocument::ReadInt32()
2244 size_t item_read = fread (&g, (size_t)4,(size_t)1, Fp);
2245 if ( item_read != 1 )
2249 dbg.Verbose(0, "gdcmDocument::ReadInt32", " File Error");
2260 * \brief skips bytes inside the source file
2261 * \warning NOT end user intended method !
2264 void gdcmDocument::SkipBytes(uint32_t nBytes)
2266 //FIXME don't dump the returned value
2267 (void)fseek(Fp, (long)nBytes, SEEK_CUR);
2271 * \brief Loads all the needed Dictionaries
2272 * \warning NOT end user intended method !
2274 void gdcmDocument::Initialise()
2276 RefPubDict = gdcmGlobal::GetDicts()->GetDefaultPubDict();
2281 * \brief Discover what the swap code is (among little endian, big endian,
2282 * bad little endian, bad big endian).
2284 * @return false when we are absolutely sure
2285 * it's neither ACR-NEMA nor DICOM
2286 * true when we hope ours assuptions are OK
2288 bool gdcmDocument::CheckSwap()
2290 // The only guaranted way of finding the swap code is to find a
2291 // group tag since we know it's length has to be of four bytes i.e.
2292 // 0x00000004. Finding the swap code in then straigthforward. Trouble
2293 // occurs when we can't find such group...
2295 uint32_t x = 4; // x : for ntohs
2296 bool net2host; // true when HostByteOrder is the same as NetworkByteOrder
2300 char deb[HEADER_LENGTH_TO_READ];
2302 // First, compare HostByteOrder and NetworkByteOrder in order to
2303 // determine if we shall need to swap bytes (i.e. the Endian type).
2304 if ( x == ntohs(x) )
2313 // The easiest case is the one of a DICOM header, since it possesses a
2314 // file preamble where it suffice to look for the string "DICM".
2315 int lgrLue = fread(deb, 1, HEADER_LENGTH_TO_READ, Fp);
2317 char *entCur = deb + 128;
2318 if( memcmp(entCur, "DICM", (size_t)4) == 0 )
2320 dbg.Verbose(1, "gdcmDocument::CheckSwap:", "looks like DICOM Version3");
2322 // Next, determine the value representation (VR). Let's skip to the
2323 // first element (0002, 0000) and check there if we find "UL"
2324 // - or "OB" if the 1st one is (0002,0001) -,
2325 // in which case we (almost) know it is explicit VR.
2326 // WARNING: if it happens to be implicit VR then what we will read
2327 // is the length of the group. If this ascii representation of this
2328 // length happens to be "UL" then we shall believe it is explicit VR.
2329 // FIXME: in order to fix the above warning, we could read the next
2330 // element value (or a couple of elements values) in order to make
2331 // sure we are not commiting a big mistake.
2332 // We need to skip :
2333 // * the 128 bytes of File Preamble (often padded with zeroes),
2334 // * the 4 bytes of "DICM" string,
2335 // * the 4 bytes of the first tag (0002, 0000),or (0002, 0001)
2336 // i.e. a total of 136 bytes.
2340 // Sometimes (see : gdcmData/icone.dcm) group 0x0002 *is* Explicit VR,
2341 // but elem 0002,0010 (Transfert Syntax) tells us the file is
2342 // *Implicit* VR. -and it is !-
2344 if( memcmp(entCur, "UL", (size_t)2) == 0 ||
2345 memcmp(entCur, "OB", (size_t)2) == 0 ||
2346 memcmp(entCur, "UI", (size_t)2) == 0 ||
2347 memcmp(entCur, "CS", (size_t)2) == 0 ) // CS, to remove later
2348 // when Write DCM *adds*
2350 // Use gdcmDocument::dicom_vr to test all the possibilities
2351 // instead of just checking for UL, OB and UI !? group 0000
2353 Filetype = gdcmExplicitVR;
2354 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2355 "explicit Value Representation");
2359 Filetype = gdcmImplicitVR;
2360 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2361 "not an explicit Value Representation");
2367 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2368 "HostByteOrder != NetworkByteOrder");
2373 dbg.Verbose(1, "gdcmDocument::CheckSwap:",
2374 "HostByteOrder = NetworkByteOrder");
2377 // Position the file position indicator at first tag (i.e.
2378 // after the file preamble and the "DICM" string).
2380 fseek (Fp, 132L, SEEK_SET);
2384 // Alas, this is not a DicomV3 file and whatever happens there is no file
2385 // preamble. We can reset the file position indicator to where the data
2386 // is (i.e. the beginning of the file).
2387 dbg.Verbose(1, "gdcmDocument::CheckSwap:", "not a DICOM Version3 file");
2390 // Our next best chance would be to be considering a 'clean' ACR/NEMA file.
2391 // By clean we mean that the length of the first tag is written down.
2392 // If this is the case and since the length of the first group HAS to be
2393 // four (bytes), then determining the proper swap code is straightforward.
2396 // We assume the array of char we are considering contains the binary
2397 // representation of a 32 bits integer. Hence the following dirty
2399 s32 = *((uint32_t *)(entCur));
2420 // We are out of luck. It is not a DicomV3 nor a 'clean' ACR/NEMA file.
2421 // It is time for despaired wild guesses.
2422 // So, let's check if this file wouldn't happen to be 'dirty' ACR/NEMA,
2423 // i.e. the 'group length' element is not present :
2425 // check the supposed to be 'group number'
2426 // 0x0002 or 0x0004 or 0x0008
2427 // to determine ' SwapCode' value .
2428 // Only 0 or 4321 will be possible
2429 // (no oportunity to check for the formerly well known
2430 // ACR-NEMA 'Bad Big Endian' or 'Bad Little Endian'
2431 // if unsuccessfull (i.e. neither 0x0002 nor 0x0200 etc -4, 8-)
2432 // the file IS NOT ACR-NEMA nor DICOM V3
2433 // Find a trick to tell it the caller...
2435 s16 = *((uint16_t *)(deb));
2452 dbg.Verbose(0, "gdcmDocument::CheckSwap:",
2453 "ACR/NEMA unfound swap info (Really hopeless !)");
2454 Filetype = gdcmUnknown;
2457 // Then the only info we have is the net2host one.
2467 * \brief Restore the unproperly loaded values i.e. the group, the element
2468 * and the dictionary entry depending on them.
2470 void gdcmDocument::SwitchSwapToBigEndian()
2472 dbg.Verbose(1, "gdcmDocument::SwitchSwapToBigEndian",
2473 "Switching to BigEndian mode.");
2474 if ( SwapCode == 0 )
2478 else if ( SwapCode == 4321 )
2482 else if ( SwapCode == 3412 )
2486 else if ( SwapCode == 2143 )
2493 * \brief during parsing, Header Elements too long are not loaded in memory
2496 void gdcmDocument::SetMaxSizeLoadEntry(long newSize)
2502 if ((uint32_t)newSize >= (uint32_t)0xffffffff )
2504 MaxSizeLoadEntry = 0xffffffff;
2507 MaxSizeLoadEntry = newSize;
2512 * \brief Header Elements too long will not be printed
2513 * \todo See comments of \ref gdcmDocument::MAX_SIZE_PRINT_ELEMENT_VALUE
2516 void gdcmDocument::SetMaxSizePrintEntry(long newSize)
2518 //DOH !! This is exactly SetMaxSizeLoadEntry FIXME FIXME
2523 if ((uint32_t)newSize >= (uint32_t)0xffffffff )
2525 MaxSizePrintEntry = 0xffffffff;
2528 MaxSizePrintEntry = newSize;
2534 * \brief Read the next tag but WITHOUT loading it's value
2535 * (read the 'Group Number', the 'Element Number',
2536 * gets the Dict Entry
2537 * gets the VR, gets the length, gets the offset value)
2538 * @return On succes the newly created DocEntry, NULL on failure.
2540 gdcmDocEntry *gdcmDocument::ReadNextDocEntry()
2542 uint16_t g = ReadInt16();
2543 uint16_t n = ReadInt16();
2547 // We reached the EOF (or an error occured) therefore
2548 // header parsing has to be considered as finished.
2551 gdcmDocEntry *newEntry = NewDocEntryByNumber(g, n);
2553 FindDocEntryVR(newEntry);
2554 FindDocEntryLength(newEntry);
2562 newEntry->SetOffset(ftell(Fp));
2569 * \brief Generate a free gdcmTagKey i.e. a gdcmTagKey that is not present
2570 * in the TagHt dictionary.
2571 * @param group The generated tag must belong to this group.
2572 * @return The element of tag with given group which is fee.
2574 uint32_t gdcmDocument::GenerateFreeTagKeyInGroup(uint16_t group)
2576 for (uint32_t elem = 0; elem < UINT32_MAX; elem++)
2578 gdcmTagKey key = gdcmDictEntry::TranslateToKey(group, elem);
2579 if (TagHT.count(key) == 0)
2588 * \brief Assuming the internal file pointer \ref gdcmDocument::fp
2589 * is placed at the beginning of a tag check whether this
2590 * tag is (TestGroup, TestElement).
2591 * \warning On success the internal file pointer \ref gdcmDocument::fp
2592 * is modified to point after the tag.
2593 * On failure (i.e. when the tag wasn't the expected tag
2594 * (TestGroup, TestElement) the internal file pointer
2595 * \ref gdcmDocument::fp is restored to it's original position.
2596 * @param TestGroup The expected group of the tag.
2597 * @param TestElement The expected Element of the tag.
2598 * @return True on success, false otherwise.
2600 bool gdcmDocument::ReadTag(uint16_t testGroup, uint16_t testElement)
2602 long positionOnEntry = ftell(Fp);
2603 long currentPosition = ftell(Fp); // On debugging purposes
2605 //// Read the Item Tag group and element, and make
2606 // sure they are what we expected:
2607 uint16_t itemTagGroup = ReadInt16();
2608 uint16_t itemTagElement = ReadInt16();
2609 if ( itemTagGroup != testGroup || itemTagElement != testElement )
2611 std::ostringstream s;
2612 s << " We should have found tag (";
2613 s << std::hex << testGroup << "," << testElement << ")" << std::endl;
2614 s << " but instead we encountered tag (";
2615 s << std::hex << itemTagGroup << "," << itemTagElement << ")"
2617 s << " at address: " << (unsigned)currentPosition << std::endl;
2618 dbg.Verbose(0, "gdcmDocument::ReadItemTagLength: wrong Item Tag found:");
2619 dbg.Verbose(0, s.str().c_str());
2620 fseek(Fp, positionOnEntry, SEEK_SET);
2628 * \brief Assuming the internal file pointer \ref gdcmDocument::fp
2629 * is placed at the beginning of a tag (TestGroup, TestElement),
2630 * read the length associated to the Tag.
2631 * \warning On success the internal file pointer \ref gdcmDocument::fp
2632 * is modified to point after the tag and it's length.
2633 * On failure (i.e. when the tag wasn't the expected tag
2634 * (TestGroup, TestElement) the internal file pointer
2635 * \ref gdcmDocument::fp is restored to it's original position.
2636 * @param TestGroup The expected group of the tag.
2637 * @param TestElement The expected Element of the tag.
2638 * @return On success returns the length associated to the tag. On failure
2641 uint32_t gdcmDocument::ReadTagLength(uint16_t testGroup, uint16_t testElement)
2643 long PositionOnEntry = ftell(Fp);
2644 (void)PositionOnEntry;
2646 if ( !ReadTag(testGroup, testElement) )
2651 //// Then read the associated Item Length
2652 long currentPosition = ftell(Fp);
2653 uint32_t itemLength = ReadInt32();
2655 std::ostringstream s;
2656 s << "Basic Item Length is: "
2657 << itemLength << std::endl;
2658 s << " at address: " << (unsigned)currentPosition << std::endl;
2659 dbg.Verbose(0, "gdcmDocument::ReadItemTagLength: ", s.str().c_str());
2665 * \brief Parse pixel data from disk for multi-fragment Jpeg/Rle files
2666 * No other way so 'skip' the Data
2669 void gdcmDocument::Parse7FE0 ()
2671 gdcmDocEntry* element = GetDocEntryByNumber(0x0002, 0x0010);
2674 // Should warn user FIXME
2678 if ( IsImplicitVRLittleEndianTransferSyntax()
2679 || IsExplicitVRLittleEndianTransferSyntax()
2680 || IsExplicitVRBigEndianTransferSyntax() /// \todo 1.2.2 ??? A verifier !
2681 || IsDeflatedExplicitVRLittleEndianTransferSyntax() )
2686 // ---------------- for Parsing : Position on begining of Jpeg/RLE Pixels
2688 //// Read the Basic Offset Table Item Tag length...
2689 uint32_t itemLength = ReadTagLength(0xfffe, 0xe000);
2691 //// ... and then read length[s] itself[themselves]. We don't use
2692 // the values read (BTW what is the purpous of those lengths ?)
2693 if ( itemLength != 0 )
2695 // BTW, what is the purpous of those length anyhow !?
2696 char* basicOffsetTableItemValue = new char[itemLength + 1];
2697 fread(basicOffsetTableItemValue, itemLength, 1, Fp);
2699 for (unsigned int i=0; i < itemLength; i += 4 )
2701 uint32_t individualLength = str2num(&basicOffsetTableItemValue[i],uint32_t);
2702 std::ostringstream s;
2703 s << " Read one length: ";
2704 s << std::hex << individualLength << std::endl;
2705 dbg.Verbose(0, "gdcmDocument::Parse7FE0: ", s.str().c_str());
2709 if ( ! IsRLELossLessTransferSyntax() )
2713 //// We then skip (not reading them) all the fragments of images:
2714 while ( (itemLength = ReadTagLength(0xfffe, 0xe000)) )
2716 SkipBytes(itemLength);
2723 long rleSegmentLength[15], fragmentLength;
2725 // While we find some items:
2726 while ( (fragmentLength = ReadTagLength(0xfffe, 0xe000)) )
2728 // Parse fragments of the current Fragment (Frame)
2729 //------------------ scanning (not reading) fragment pixels
2730 uint32_t nbRleSegments = ReadInt32();
2732 //// Reading RLE Segments Offset Table
2733 uint32_t rleSegmentOffsetTable[15];
2734 for(int k=1; k<=15; k++)
2736 ftellRes = ftell(Fp);
2737 rleSegmentOffsetTable[k] = ReadInt32();
2740 // skipping (not reading) RLE Segments
2741 if ( nbRleSegments > 1)
2743 for(unsigned int k = 1; k <= nbRleSegments-1; k++)
2745 rleSegmentLength[k] = rleSegmentOffsetTable[k+1]
2746 - rleSegmentOffsetTable[k];
2747 ftellRes = ftell(Fp);
2748 SkipBytes(rleSegmentLength[k]);
2752 rleSegmentLength[nbRleSegments] = fragmentLength
2753 - rleSegmentOffsetTable[nbRleSegments];
2754 ftellRes = ftell(Fp);
2755 SkipBytes(rleSegmentLength[nbRleSegments]);
2758 // Make sure that at the end of the item we encounter a 'Sequence
2760 if ( ! ReadTag(0xfffe, 0xe0dd) )
2762 dbg.Verbose(0, "gdcmDocument::Parse7FE0: no sequence delimiter item");
2763 dbg.Verbose(0, " at end of RLE item sequence");
2771 * \brief Compares two documents, according to \ref gdcmDicomDir rules
2772 * \warning Does NOT work with ACR-NEMA files
2773 * \todo Find a trick to solve the pb (use RET fields ?)
2775 * @return true if 'smaller'
2777 bool gdcmDocument::operator<(gdcmDocument &document)
2780 std::string s1 = this->GetEntryByNumber(0x0010,0x0010);
2781 std::string s2 = document.GetEntryByNumber(0x0010,0x0010);
2793 s1 = this->GetEntryByNumber(0x0010,0x0020);
2794 s2 = document.GetEntryByNumber(0x0010,0x0020);
2805 // Study Instance UID
2806 s1 = this->GetEntryByNumber(0x0020,0x000d);
2807 s2 = document.GetEntryByNumber(0x0020,0x000d);
2818 // Serie Instance UID
2819 s1 = this->GetEntryByNumber(0x0020,0x000e);
2820 s2 = document.GetEntryByNumber(0x0020,0x000e);
2836 //-----------------------------------------------------------------------------